AAS 99-181

A New Algorithm for Star Pattern Recognition

D. Mortari

Universita' degli Studi "La Sapienza" di Roma

Abstract

A new star identification, based on the fact that any star direction C can be expressed as a linear combination of two given directions (A and B) and their cross product (AxB), is presented. Stars observed are expressed as a function of two reference stars, A and B. This relationship, within the instrument precision, can also be written in the inertial reference system. Using the K-vector technique, a subset of inertial star pair matching the observed A and B, is selected. For each entry two inertial directions (V' and V") for C are evaluated. The problem becomes to identify the inertial star closest to V' (and V"), which is solved again using the K-vector technique applied to the star components. The Star is identified if there is only one star which satisfies the requirements. This procedure, which is then repeated for any other star observed, yields the complete star pattern recognition. The described method is able toidentify and discard spikes and it is compared with the Search-Less Algorithm. Plots show both the algorithm speed and robustness.